PROJECT SUMMARY ? Project 1 High grade serous ovarian cancer (HGSOC), the most common and most lethal subtype, is largely a disease of genomic instability. The tumor suppressor TP53 is inactivated in over 95% of cases; and the Fanconi anem- ia/homologous recombination (FA/HR) pathway is mutationally inactivated in 30-40%, leading to further gen- omic instability. Inhibitors of the DNA repair protein poly(ADP-ribose) polymerase (PARP) have exhibited prom- ising activity in HR-deficient preclinical models, leading to extensive efforts to develop PARP inhibitors for HGSOC and other HR-deficient cancers. Although phase II clinical trials have shown that PARP inhibitors pro- duce i) objective response rates of 30-45% in BRCA1 or BRCA2 (BRCA1/2) mutation carriers with platinum- sensitive relapsed ovarian cancer and ii) substantial prolongation of progression-free survival in the mainten- ance setting after response of BRCA1- or BRCA2-mutant HGSOCs to platinum-containing therapy, the FDA has highlighted the need to better identify ovarian cancers that are most likely to respond to PARP inhibitors. Building on results generated during the previous funding period, the present studies are designed to better understand why some HR-deficient ovarian cancers respond to PARP inhibitors and others do not. Our preliminary results show that the cytotoxic effects of PARP inhibitors in HR-deficient ovarian cancer reflect activation of nonhomologous end-joining (NHEJ), an error-prone repair pathway, rather than interruption of single-strand break repair as originally postulated by others. Importantly, inhibition of the NHEJ-associated kinase DNA-PK or downregulation of any of a number of NHEJ proteins simultaneously inhibits this error-prone repair and diminishes the cytotoxic effects of PARP inhibitors. This new understanding of PARP inhibitor action suggests the hypothesis that ovarian cancers will respond to PARP inhibitors only if HR is defective and the NHEJ pathway remains intact. Consistent with this hypothesis, our further studies in a BRCA2-mutant preclinical ovarian cancer model have revealed that selection for PARP inhibitor resistance results in either downregulation of NHEJ proteins or overexpression of Rad51, an HR protein downstream of BRCA2, sug- gesting that disabling NHEJ or restoring HR by means other than BRCA1/2 mutations confers PARP inhibitor resistance. To build on these findings we now propose to: i) determine how Rad51 overexpression contributes to PARP inhibitor resistance, ii) identify the changes that confer PARP inhibitor resistance in HR-deficient patient-derived ovarian cancer xenografts in vivo and iii) develop a multi-parameter classifier that includes assays of Rad51 and NHEJ protein expression, sequencing of repair genes and measurements of genomic scarring, to predict responses to the PARP inhibitor rucaparib in a large multicenter phase II trial in patients with relapsed ovarian cancer. Impact: Collectively, these studies will provide new insight into mechanisms of PARP inhibitor resistance and simultaneously test the concept that a multifaceted assessment of repair path- way integrity will identify ovarian cancer patients most likely to benefit from this promising new class of drugs.